Early goal-directed therapy Where to from here? Rinaldo Bellomo ANZIC Research Centre Melbourne, Australia

Transcription

1 Early goal-directed therapy Where to from here? Rinaldo Bellomo ANZIC Research Centre Melbourne, Australia

2 Early goal-directed therapy in septic shock 2001 Proof-of-concept EGDT trial published NEJM 16% ARR in-hospital mortality 46.5% vs. 30.5%, P = , 2008, 2012 SSC guidelines recommend EGDT for first 6 hours For all patients with severe sepsis Grade 1 (strong) C (low-quality) Mortality benefit in > 20 non-randomised studies of bundled care Rivers E. New Engl J Med 2001:345:1368 EGDT protocol

3 In large, prospective, multi-centre, Phase III RCTs EGDT vs. standard care in patients presenting to ED now published Protocolised Care for Early Septic Shock United States University of Pittsburgh, Derek Angus Australasian Resuscitation In Sepsis Evaluation Australia, NZ, Hong Kong, Finland, Ireland ARISE Investigators, ANZICS CTG, ACEM CTG Protocolised Management In Sepsis England ICNARC, Kathy Rowan

4 Single research question in all three trials HYPOTHESIS EGDT provides early recognition of systemic hypoperfusion in early septic shock Resolution of global tissue hypoxia would improve outcome AIM Determine whether providing EGDT, compared to standard care, reduced mortality in patients presenting to the ED of hospitals in Australasia (ARISE), United States (ProCESS) and England (ProMISE) with early septic shock

5 ProCESS, ARISE and ProMISe inclusion criteria 1) 18 years or over 2) Suspected or confirmed infection 3) Two or more SIRS criteria 4) Refractory hypotension OR o SBP < 90 mm Hg or MAP < 65 mm Hg after 1000 ml i.v. fluid within 60 mins 5) Hypoperfusion o Blood lactate 4.0 mmol/l All criteria met within 6 hours of ED presentation (12 hours for ProCESS) Randomisation within 2 hours of fulfilling final inclusion criterion Commence first dose i.v. antimicrobial prior to randomisation ( ARISE, ProMISe only)

6 Study treatments Prior informed or delayed consent from patient or legal surrogate Randomised to receive either usual care or EGDT for 6 hours (stratified by site) USUAL CARE GROUP o Investigations, monitoring and all treatments determined by treating clinicians EGDT GROUP o Dedicated study team (doctor, nurse, coordinator) trained in EGDT delivery o Care providers and location of delivery dependent on local resources ED-based, ICU-based and ED/ICU hybrid models ProCESS third arm protocolised standard therapy (1:1:1 randomisation)

7 ProCESS, ARISE and ProMISe EGDT algorithm Arterial catheter and ScvO 2 -CVC inserted within 1 hour randomisation CVP 8 (SV) or 12 (MV) mm Hg o 500 ml i.v. fluid bolus MAP mm Hg o Vasoactive agent ScvO 2 70% o Red-cell transfusion (Hct < 30%) o Dobutamine

8 Why did we need three trials? Why were there 3 separate, similar trials of EGDT with a single research question? Why did 3 government funding bodies agree to fund 3 concurrent trials in own regions? NIH 9 million US dollars NHMRC 2 million AUS dollars NIHR 6 million pounds Why would the NEJM publish 3 trials of the same intervention within 1 year?

9 Generalisability of the evidence EGDT bundle is a process of care intervention Evaluation and widespread delivery of EGDT is contextdependent Not the same as testing drug A vs. drug B Could the same intervention be faithfully delivered in different regions? Could results one health care setting be uniformly applied across all health systems? Different health care systems and organisational structures Different technological, financial and man-power resources Different patient populations, case-mix, therapies and usual care

10 May 1, 2014 ProCESS trial Conducted in 31 USA tertiary care centres 1341 patients within 12 hours of presenting to ED with severe sepsis EGDT (n=439) and protocolised standard Rx (n=446) vs. usual care (n=456) (1 0 hypothesis) 19.5% vs. 18.9% hospital mortality censored at 60 days; RR 1.04 ( ) P = 0.83

11 October 1, 2014 ARISE trial

12 ARISE participating centres Conducted in 51 sites o Australia 41 o Finland 2 o Hong Kong 3 o New Zealand 3 o Republic of Ireland 1 Diversity of health care settings o Tertiary metropolitan 31 o Non-tertiary metropolitan 13 o Non-tertiary rural 6 15,000 to 200,000 ED presentations per annum

13 % patients Therapies delivered between 0 and 6 hours Intravenous fluids EGDT Usual care P value 0 to 6 hours 1964 ± 1415 ml 1713 ± 1401 ml < P < for all Vasopressor RBC Dobutamine EGDT

14 EGDT vs. usual care attainment of goals SpO 2 CVP P = 0.6 P = 0.23 MAP ScvO 2 (EGDT only) At 6 hours EGDT 76.5 ± 10.8 vs. usual care 75.3 ± 11.4, P = 0.04

15 Primary outcome No difference between EGDT and usual care in 90-day all-cause mortality o EGDT 18.6 % (147/792) vs. usual care 18.8% (150/796) (P = 0.90) o Absolute risk difference -0.3% (95% CI -4.1% to 3.6%) o Relative risk 0.98 (95% CI 0.80 to 1.21) Survival time to day 90 P = 0.82

16 Secondary and tertiary outcomes EGDT patients shorter ED length of stay post-randomisation o EGDT 1.4 ( ) vs. usual care 2.0 ( ) hours (P < 0.001) More EGDT patients admitted to ICU direct from ED o EGDT 87.0% vs. usual care 76.9% patients (P < 0.001) o ICU admission to facilitate delivery of the EGDT algorithm o 14.0% EGDT patients (111) only admitted to enable study intervention More EGDT patients overall received a vasopressor infusion o EGDT 76.3% vs. usual care 65.8% (P < 0.001) o Duration of administration not different EGDT 29.4 ( ) vs ( ) hours (P = 0.24)

17 Post-ProCESS and ARISE YES - EGDT results are applicable outside of the US (i.e. Australasia ) YES - EGDT can be delivered in 3 ry vs. non-3 ry and metropolitan vs. rural sites? MAYBE - results obtained in one health care setting/organisational structure be uniformly applied across all other health care systems throughout the world? What about the UK? Do the Surviving Sepsis Campaign guidelines need to be revised?

18 Surviving sepsis campaign response Based on release of ProCESS and ARISE trials, the SSC Executive Committee states: Results of the ProCESS and ARISE trials have not demonstrated any adverse outcomes in the groups that utilized CVP and ScvO2 as end-points for resuscitation. Therefore, no harm exists in keeping the current SSC bundles intact as recent published evidence (observational, before-after) shows significant mortality reduction using the existing SSC bundles SSC guidelines committee will immediately review the evidence and assess whether the guidelines need to be updated now

19 March 17, 2014 ProMISe trial Conducted in 56 tertiary care, regional and rural centres in England 1260 patients within 6 hours of presenting to ED with severe sepsis EGDT (n=630) and vs. usual care (n=630) 29.5% vs. 29.2% 90-day mortality (RR 1.01 ( ) P = % vs 18.8% ARISE and 31.9% and 33.7% ProCESS

20 Updated bundles in response to new evidence No suggestion of harm with use of CVC in any trial Published evidence shows significant mortality reduction using original SSC bundles Weaknesses in all current studies BUT.with publication of 3 trials that do not demonstrate superiority of required use of CVC to monitor CVP and ScvO 2 in all patients with septic shock who have received timely antibiotics and fluid resuscitation or with lactate > 4 mmol/l, SSC Executive committee has revised improvement bundle..

21 Within 6 hrs of presentation Apply vasopressors to maintain MAP > 65 mm Hg If persistent hypotension, reassess volume status and tissue perfusion Repeated focussed exam by licensed independent practitioner Vital signs Cardiopulmonary Capillary refill Pulse Skin findings OR 2 of the following CVP ScvO 2 Bedside cardiovascular ultrasound Dynamic assessment using passive leg raise or fluid challenge Remeasure lactate if initial elevated

22 We told you so.

23

24 Systematic review and meta-analysis EGDT trials 1 January 2000 to 15 March 2015

25 Included trials EGDT vs. non-egdt resuscitation 6 usual care; 3 lactate clearance; 1 PiCCO 6 multi-centre; 5 single-centre Author Year Region N o sites Population Source Control(s) N o patients 1 0 outcome PRIMARY OBJECTIVE (presentation to the ED) Rivers et al USA 1 Adult ED Usual care 263 In-hospital Jones et al USA 3 Adult ED Lactate clearance 300 In-hospital ProCESS Investigators 2014 USA 31 Adult ED Usual care, protocol-based standard Rx 1341 In-hospital ARISE Investigators 2014 Australasia 51 Adult ED Usual care day ProMISe Investigators 2015 UK 56 Adult ED Usual care day SECONDARY OBJECTIVE (irrespective of presenting source) Wang et al China 1 Adult Unknown Usual care day De Oliviera etal Brazil 2 Paediatric ED, ward, ICU ACCM/PALS guidelines day EGDT collaborative 2010 China 8 Adult Unknown Usual care day Tian etal China 1 Adult Unknown 10% or 30% lactate clearance day Yu et al China 1 Adult Unknown Lactate clearance 10% day Lu et al China 1 Adult Unknown PiCCO-guided resuscitation 82 In-hospital TOTAL PATIENTS 5407

26 Independent risk of bias assessment Author Random Allocation Blinding participants Blinding outcome Incomplete Selective Other sequence concealment personnel assessment outcome data reporting sources bias generation PRIMARY OBJECTIVE Rivers et al. Low Low High Low Low Unclear Low Jones et al. Low Low High Low Low Unclear Low ProCESS Investigators Low Low High Low Low Low High ARISE Investigators Low Low High Low Low Low Low ProMISe Investigators Low Low High Low Low Low Low SECONDARY OBJECTIVE Wang et al. Unclear Unclear High Unclear Low Unclear Unclear De Oliveira et al. Low Low High Low Low Low High EGDT collaborative Low Unclear High Unclear Low Unclear Unclear Tian et al. Low Unclear High Unclear High Unclear Unclear Yu et al. Low Unclear High Unclear Low Unclear Unclear Lu etal. Low Unclear High Unclear Low Unclear Unclear

27 Patient characteristics Author N o patients APACHE II score Hypotension, % Lactate, mmol/l 1 0 mortality, % EGDT Control EGDT Control EGDT Control EGDT Control EGDT Control PRIMARY OBJECTIVE Rivers et al ± ± ± ± Jones et al NA NA ± ± ProCESS Investigators ± ± ± ± ARISE Investigators ± ± ± ± ProMISe Investigators ± ± ± ± SECONDARY OBJECTIVE Wang et al ± 7 27 ± NA NA De Oliviera et al NA NA ( ) 1.2 ( ) EGDT collab ve ± ± 6.5 NA NA NA NA Tian et al ± ± NA NA Yu et al ± ± NA 3.4 ± Lu et al ± ± ± ± TOTAL

28 Interventions delivered Author Fluids, ml Vasopressor, % Dobutamine, % Blood transfusion, % Time to first antimicrobial (min), EGDT Control EGDT Control EGDT Control EGDT Control EGDT Control PRIMARY OBJECTIVE Rivers et al 4981 ± 3499 ± NA NA Jones et al 4300 ± 4500 ± (66-170) 115 (62-180) ProCESS Investigators 2805 ± 2783 ± NA NA ARISE Investigators 1964 ± 1713 ± (38-114) 67 (39-110) ProMISe 2226 ± 2022 ± NA NA Investigators SECONDARY OBJECTIVE Wang et al 4895 ± NA NA NA NA NA NA De Oliviera et al NA a NA a NA NA

29 Interventions delivered Author Fluids, ml Vasopressor, % Dobutamine, % Blood transfusion, % Time to first antimicrobial (min) EGDT Control EGDT Control EGDT Control EGDT Control EGDT Control PRIMARY OBJECTIVE Rivers et al 4981 ± 3499 ± NA NA Jones et al 4300 ± 4500 ± (66-170) 115 (62-180) ProCESS Investigators 2805 ± 2783 ± NA NA ARISE Investigators 1964 ± 1713 ± (38-114) 67 (39-110) ProMISe 2226 ± 2022 ± NA NA Investigators SECONDARY OBJECTIVE Wang et al 4895 ± NA NA NA NA NA NA De Oliviera et al NA a NA a NA NA

30 Interventions delivered Author Fluids, ml Vasopressor, % Dobutamine, % Blood transfusion, % Time to first antimicrobial (min), EGDT Control EGDT Control EGDT Control EGDT Control EGDT Control PRIMARY Rivers et al 4981 ± 3499 ± NA NA Jones et al 4300 ± 4500 ± (66-170) 115 (62-180) ProCESS 2805 ± 2783 ± NA NA Investigators ARISE 1964 ± 1713 ± (38-114) 67 (39-110) ProMISe 2226 ± 2022 ± NA NA

31 No effect on primary study mortality outcome PRIMARY OBJECTIVE: ED-BASED STUDIES (n=5) Study Events, Events, % ID OR (95% CI) EGDT comparator Weight Rivers et al. (2001) 0.52 (0.31, 0.86) 38/130 59/ Jones et al. (2010) 1.47 (0.82, 2.60) 34/150 25/ ProCESS Investigators (2014) 1.17 (0.88, 1.55) 92/ / ARISE Investigators (2014) 0.98 (0.76, 1.26) 147/ / ProMISe Investigators (2015) 1.02 (0.80, 1.30) 184/ /620 P= Overall (I-squared = 56.7%, p = 0.055) 1.01 (0.88, 1.16) 495/ / Favours EGDT Favours comparator Fixed-effect model Random-effects model did not change results Inclusion of only usual care as control did not change results (i.e excluding Jones and PrOCESS protocolised standard therapy)

32 No effect on primary study mortality outcome SECONDARY OBJECTIVE: INDEPENDENT OF PRESENTING SOURCE (n=11) Study ID OR (95% CI) Events, EGDT Events, control % Weight Source: ED Rivers et al. (2001) Jones et al. (2010) ProCESS Investigators (2014) ARISE Investigators (2014) ProMISe Investigators (2015) Subtotal (I-squared = 56.7%, p = 0.055). Source: other/unclear Wang et al. (2006) de Oliveira et al. (2008) EGDT Collaborative (2010) Tian et al. (2012) Yu et al. (2013) Lu et al. (2014) Subtotal (I-squared = 72.2%, p = 0.003). Overall (I-squared = 70.7%, p = 0.000) 0.52 (0.31, 0.86) 1.47 (0.82, 2.60) 1.17 (0.88, 1.55) 0.98 (0.76, 1.26) 1.02 (0.80, 1.30) 1.01 (0.88, 1.16) 0.48 (0.11, 2.11) 0.21 (0.07, 0.57) 0.46 (0.28, 0.74) 3.55 (1.15, 10.99) 1.56 (0.42, 5.78) 1.06 (0.34, 3.35) 0.61 (0.43, 0.86) 0.94 (0.82, 1.07) 38/130 34/150 92/ / / /2134 4/16 6/51 41/163 12/19 7/25 7/40 77/ / /133 25/ / / / /2601 7/17 20/51 64/151 14/43 5/25 7/42 117/ / P= Favours EGDT Favours control Fixed-effect model No evidence publication bias across the 11 included studies Random-effects model did not change results Heterogeneity increased with inclusion of non-ed based studies (70.7% vs. 56.7%) No interaction between presenting source and treatment effect

33 No effect on 90-day mortality outcome Study Events, Events, % ID OR (95% CI) EGDT comparator Weight ProCESS Investigators (2014) 0.98 (0.76, 1.26) 129/ / ARISE Investigators (2014) 0.98 (0.76, 1.26) 147/ / ProMISe Investigators (2015) 1.02 (0.80, 1.30) 184/ /620 P= Overall (I-squared = 0.0%, p = 0.974) 0.99 (0.86, 1.15) 460/ / Favours EGDT Favours comparator Fixed-effect model Random-effects model did not change results Heterogeneity 0.0% 28-day and hospital mortality results not different

34 Increased receipt of vasopressors Study Events, Events, % ID OR (95% CI) EGDT control Weight Rivers et al. (2001) 0.56 (0.34, 0.92) 48/130 68/ Jones et al. (2010) 1.29 (0.79, 2.10) 108/ / ProCESS Investigators (2014) 1.06 (0.84, 1.34) 269/ / ARISE Investigators (2014) 1.67 (1.34, 2.08) 605/ / ProMISe Investigators (2015) 1.25 (1.00, 1.57) 377/623 P< / Overall (I-squared = 78.9%, p = 0.001) 1.25 (1.10, 1.41) 1407/ / Favours EGDT Favours control No effect of EGDT on receipt or duration of ventilation or RRT

35 Increased ICU utilisation Study Events, Events, % ID OR (95% CI) EGDT comparator Weight ICU ADMISSION ProCESS Investigators (2014) 1.75 (1.19, 2.56) 401/ / ARISE Investigators (2014) 2.21 (1.62, 3.02) 725/ / ProMISe Investigators (2015) 2.54 (1.87, 3.43) 551/ / Jones et al. (2010) (Excluded) 150/ / Overall (I-squared = 11.6%, p = 0.323) 2.19 (1.82, 2.65) 1827/ /2474 P< Favours EGDT Favours comparator ICU LOS Study N, mean N, mean % ID WMD (95% CI) (SD); EGDT (SD); comparator Weight Jones et al. (2010) (-2.10, 1.50) 150, 5.6 (7.39) 150, 5.9 (8.46) 6.31 ProCESS Investigators (2014) 0.20 (-0.57, 0.97) 401, 5.1 (6.3) P= , 4.9 (6.5) ARISE Investigators (2014) 0.10 (-0.59, 0.79) 725, 4.9 (7.2) 661, 4.8 (5.9) ProMISe Investigators (2015) (-1.82, 0.42) 549, 5.7 (8.1) 466, 6.4 (9.8) Overall (I-squared = 0.0%, p = 0.584) (-0.47, 0.43) Favours EGDT Favours comparator

36 Limitations of a trial-level meta-analysis International and local variation in healthcare services Hospital type (tertiary vs metro vs regional) Number ED presentations Threshold for ICU admission Effect of individual patient confounders age, sex, comorbidities, source infection Severity illness, presence shock or hyperlactataemia at baseline, ScvO 2 at insertion Nature of usual care across sites across sites and across studies How EDDT delivered across sites and across studies

37 Individual patient data meta-analysis (IPDMA) ProCESS, ARISE and ProMISe conducted contemporaneously and independently All 3 RCTs conducted in a collaborative and harmonised fashion Pre-planned, pre-specified pooled analysis of the individual data from all 3 trials Examine broad external validity of EDT Conduct appropriately powered sub-group analyses in clinically important groups Identify patient groups most likely to benefit e.g. shock, low ScvO2 Identify therapies most likely to benefit e.g. fluids vs. vasopressors, type/timing Power to examine effect on secondary outcomes e.g. LOS, LOV, receipt vasopressors

38 Things one CAN ALREADY EASILY SEE

39 WHAT IS VISIBLE AT A GLANCE..

40 Conclusions In 2001 Rivers reported EGDT reduced mortality in ED patients with septic shock Surviving Sepsis Campaign consistently recommended EGDT for next decade Three large, multi-centre, RCTs have now confirmed no survival benefit SSC guidelines revised guidelines But ScvO 2 still in the package as an option EGDT implementation costly Resources increased ICU admission Equipment - catheters and monitors In an era of evidence-based medicine EGDT has NO ROLE in resuscitation in 2015

41 Where to from here? Use IPDMA (n=4,200) to explore signals regarding specific early interventions and associations with outcome New approaches to resuscitation to be tested (phase II RCT of early peripheral vein vasopressor therapy to start in ANZ in 2015) with early use of echocardiography to identify suspected low CO Deeper questioning of what it is that we try to achieve with resuscitation in sepsis

42

43 Key questions we need to address Do lots of fluids during resuscitation really help patients survive more? Would vasopressors achieve the same physiological changes faster/better? What if increasing the delivery of cytokine/mediators filled blood to organs promoted organ injury? What if organs are malfunctioning due to inflammatory injury and not due inadequate flow/o2? What if sepsis associated hypotension has evolved over millions of years to protect us?

44 More Key Questions Does fluid bolus resuscitation really increase vital organ blood flow? If so, by how much and for how long? Does it increase CO? If so, how much and for how long? Should we increase CO at all? What is the physiological price we pay for interventions (organ edema, worse gas exchange, acid-base disorders)? Is it worth it? Sadly WE DO NOT HAVE DATA! We do not know organ blood flow at sepsis presentation or CO or the effect of fluid bolus resuscitation on these variable We certainly do not if the gain (if present) is worth the price in terms of patient-centered outcomes

45 What next? EGDT is dead & fluids-based resuscitation in sepsis is under attack Early vasopressor therapy is now being reviewed Inotropes, blood, oxygen, fluids are all open to challenge But what is the goal of resuscitation? Is sepsis a disease of dys-immunity or organ dysoxia or both. Is organ injury due to immunological mechanisms, hemodynamic mechanisms or tissue oxygen related problems or all three? If so, what matters the most of these? The truth is that at a deeper level when resuscitating septic patients.outside of the obvious WE DO NOT REALLY KNOW WHAT WE ARE DOING AND WHY!